test(api): migrate App and Agent runner coverage to service adapters

api/services/agent/chat/base_runner.py

@@ -412,9 +412,9 @@ def organize_agent_history(
                         else:
                             tool_responses = dict.fromkeys(tool_names, observation_payload)
 
-                        for tool in tool_names:
-                            # generate a uuid for tool call
-                            tool_call_id = str(uuid.uuid4())
+                        for index, tool in enumerate(tool_names):
+                            # Keep IDs stable across turns so replayed history can use prompt caching.
+                            tool_call_id = str(uuid.uuid5(uuid.NAMESPACE_OID, f"{agent_thought.id}:{index}:{tool}"))
                             tool_calls.append(
                                 AssistantPromptMessage.ToolCall(
                                     id=tool_call_id,

api/services/workflow/execution/adapters/response_converter.py

@@ -275,6 +275,7 @@ def workflow_start_to_stream_response(
         run_id = self._ensure_workflow_run_id(workflow_run_id)
         started_at = naive_utc_now()
         self._workflow_started_at = started_at
+        self._application_generate_entity.workflow_execution_timing.start()
         if reason == WorkflowStartReason.RESUMPTION:
             self._node_snapshots.update(
                 (NodeExecutionId(snapshot.execution_id), snapshot) for snapshot in node_execution_snapshots
@@ -310,7 +311,7 @@ def workflow_finish_to_stream_response(
             )
 
         finished_at = naive_utc_now()
-        elapsed_time = (finished_at - started_at).total_seconds()
+        elapsed_time = self._application_generate_entity.workflow_execution_timing.finish()
 
         outputs_mapping = graph_runtime_state.outputs or {}
         encoded_outputs = WorkflowRuntimeTypeConverter().to_json_encodable(outputs_mapping)
@@ -363,8 +364,7 @@ def workflow_pause_to_stream_response(
             raise ValueError(
                 "workflow_pause_to_stream_response called before workflow_start_to_stream_response",
             )
-        paused_at = naive_utc_now()
-        elapsed_time = (paused_at - started_at).total_seconds()
+        elapsed_time = self._application_generate_entity.workflow_execution_timing.finish()
         encoded_outputs = self._encode_outputs(event.outputs) or {}
         if self._application_generate_entity.invoke_from == InvokeFrom.SERVICE_API:
             encoded_outputs = {}

api/services/workflow/execution/adapters/workflow/app_generator.py

@@ -42,7 +42,6 @@
 from models.enums import WorkflowRunTriggeredFrom
 from models.model import App, EndUser
 from models.workflow import WorkflowNodeExecutionTriggeredFrom
-from repositories.entities.workflow_execution import WorkflowExecutionIdentity
 from services.app.file_input_defaults import grant_file_default_access
 from services.app.generation.input_adapter import AppInputAdapter
 from services.workflow.contracts import WorkflowSnapshot
@@ -60,6 +59,7 @@
 
 if TYPE_CHECKING:
     from services.workflow.execution.contracts import WorkflowRunDriver
+    from services.workflow_execution_identity import WorkflowExecutionIdentity
 
 logger = logging.getLogger(__name__)
 

api/tests/unit_tests/core/agent/conftest.py

@@ -1,5 +1,9 @@
+from unittest.mock import MagicMock, create_autospec
+
 import pytest
 
+from services.agent.chat.ports import AgentToolInvoker
+
 
 class DummyTool:
     def __init__(self, name):
@@ -78,3 +82,8 @@ def _factory(**kwargs):
         return DummyScratchpadUnit(**kwargs)
 
     return _factory
+
+
+@pytest.fixture
+def agent_tool_invoker() -> MagicMock:
+    return create_autospec(AgentToolInvoker, instance=True, spec_set=True)